一个带索引联合的：

df = pd.DataFrame(columns=range(16)).add_prefix('col')

mid = len(df.columns)//2
idx = df.columns[:mid].union(df.columns[[-1]]
                      .union(df.columns[mid:-1],sort=False),sort=False)
out = df[idx]

Empty DataFrame
Columns: [col0, col1, col2, col3, col4, col5, col6, col7, 
         col15, col8, col9, col10, col11, col12, col13, col14]
Index: []

pop + insert

import pandas as pd
df = pd.DataFrame({f'c{i}': [i] for i in range(1,16)})

cname = df.columns[-1]
df.insert(int(len(df.columns)/2), cname, df.pop(cname))

print(df)
#   c1  c2  c3  c4  c5  c6  c7  c15  c8  c9  c10  c11  c12  c13  c14
#0   1   2   3   4   5   6   7   15   8   9   10   11   12   13   14

np.r_

创建一个用于切片的.iloc索引列表。有一些数学问题需要处理列数是偶数还是奇数。

N = int(len(df.columns)/2)
r = len(df.columns)%2
idx = np.r_[0:N, -1, -N-r:-1]
#array([ 0,  1,  2,  3,  4,  5,  6, -1, -8, -7, -6, -5, -4, -3, -2])

df = df.iloc[:, idx]
#   c1  c2  c3  c4  c5  c6  c7  c15  c8  c9  c10  c11  c12  c13  c14
#0   1   2   3   4   5   6   7   15   8   9   10   11   12   13   14